Segfault when building in docker-container builder, but not docker builder, on aarch64 (native)

[thespad]

Contributing guidelines

• I've read the contributing guidelines and wholeheartedly agree

I've found a bug and checked that ...

• ... the documentation does not mention anything about my problem
• ... there are no open or closed issues that are related to my problem

Description

This is a difficult issue to break down because I haven't been able to isolate the specific combination of elements that are triggering the crash.

We have an image that we are building for arrach64 on native aarch64 hardware - https://github.com/linuxserver/docker-mastodon/blob/main/Dockerfile.aarch64

Building with the default (docker) builder is fine, building with a docker-container builder causes a segfault within Ruby during asset precompilation.

The same issue does not occur buliding for amd64 on native amd64 hardware with either buildx driver.

I appreciate there's a good chance you're going to throw this back at Ruby/the upstream Gem project but nonetheless I'm out of ideas of how to further troubleshoot this and it only seems to be happening with the docker-container driver and I've not been able to find any similar issues reported anywhere that aren't qemu-related.

Expected behaviour

No segfault

Actual behaviour

Segfault

Buildx version

github.com/docker/buildx v0.17.1 257815a

Docker info

Client: Docker Engine - Community
 Version:    27.3.1
 Context:    default
 Debug Mode: false
 Plugins:
  buildx: Docker Buildx (Docker Inc.)
    Version:  v0.17.1
    Path:     /usr/libexec/docker/cli-plugins/docker-buildx
  compose: Docker Compose (Docker Inc.)
    Version:  v2.29.7
    Path:     /usr/libexec/docker/cli-plugins/docker-compose

Server:
 Containers: 10
  Running: 10
  Paused: 0
  Stopped: 0
 Images: 13
 Server Version: 27.3.1
 Storage Driver: overlay2
  Backing Filesystem: extfs
  Supports d_type: true
  Using metacopy: false
  Native Overlay Diff: true
  userxattr: false
 Logging Driver: json-file
 Cgroup Driver: systemd
 Cgroup Version: 2
 Plugins:
  Volume: local
  Network: bridge host ipvlan macvlan null overlay
  Log: awslogs fluentd gcplogs gelf journald json-file local splunk syslog
 Swarm: inactive
 Runtimes: io.containerd.runc.v2 runc
 Default Runtime: runc
 Init Binary: docker-init
 containerd version: 57f17b0a6295a39009d861b89e3b3b87b005ca27
 runc version: v1.1.14-0-g2c9f560
 init version: de40ad0
 Security Options:
  apparmor
  seccomp
   Profile: builtin
  cgroupns
 Kernel Version: 6.8.0-1013-oracle
 Operating System: Ubuntu 24.04.1 LTS
 OSType: linux
 Architecture: aarch64
 CPUs: 4
 Total Memory: 23.42GiB
 Name: arm-unit-01
 ID: NGGZ:6CGK:4TBF:PJLR:3IOG:QKYA:J2LX:XHR3:X6YZ:M376:V3QX:TBAP
 Docker Root Dir: /var/lib/docker
 Debug Mode: false
 Experimental: false
 Insecure Registries:
  127.0.0.0/8
 Live Restore Enabled: true
 Default Address Pools:
   Base: 172.20.0.0/16, Size: 24

Builders list

NAME/NODE        DRIVER/ENDPOINT                   STATUS    BUILDKIT   PLATFORMS
container        docker-container
 \_ container0    \_ unix:///var/run/docker.sock   running   v0.17.2    linux/arm64, linux/arm/v7, linux/arm/v6
default*         docker
 \_ default       \_ default                       running   v0.16.0    linux/arm64, linux/arm/v7, linux/arm/v6

Configuration

buildx instance created with: docker buildx create --driver docker-container --name container --bootstrap

Build invoked with docker buildx build -t mastodon -f Dockerfile.aarch64 --builder container .

Dockerfile: https://github.com/linuxserver/docker-mastodon/blob/main/Dockerfile.aarch64

Specific command causing the segfault: https://github.com/linuxserver/docker-mastodon/blob/main/Dockerfile.aarch64#L67

Build logs:
log-preamble.txt
segfault.txt

Build logs

Additional info

Have also tried using the same buildkit version (moby/buildkit:v0.16.0) as the docker driver for the container driver but the same issue is present.

Have tried to rebuild google-protobuf with force_ruby_platform to recompile it at build time without native extensions but 3.25.5 has a bug that causes it to fail and I can't really start breaking dependency chains in the application to try and use a newer version.